Open-source AI Security Operations Center: alert fusion, LLM-agent triage, MITRE ATT&CK investigation, and a replayable decision ledger for every agent step. Self-hostable, runs with no API keys, MIT licensed. Ships an MCP server for Claude, Cursor and Continue.
Python agentic-aiai-securityalert-triageblue-teamcybersecuritydetection-engineeringincident-responselanggraphllm-agentsmcpmitre-attackmodel-context-protocolpurple-teamsecurity-operationsself-hostedsiemsoarsocthreat-detectionthreat-intelligence
fix(agents): point threat-actor attribution at threatintel:8005 (not :8083) (#327)
Aadit Singal6aea406687fa
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex 618b62fb2..284625d03 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -7,6 +7,20 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0## [Unreleased]+### Fixed++- **Threat-actor attribution could never connect.** The investigation+ agent defaulted `AISOC_THREATINTEL_URL` to `http://threatintel:8083`,+ but the `threatintel` service binds port **8005** (Dockerfile,+ `docker-compose.yml`, and the README service table). Every+ `POST /api/v1/actors/attribute` call from+ `services/agents/app/agents/investigation_agent.py` therefore targeted a+ port nothing listens on and failed (soft-handled, so attribution silently+ degraded). Corrected the default to `http://threatintel:8005`, fixed the+ matching `docs/threat-actor-attribution.md` references (and a stale+ `AISOC_ATTRIBUTION_TIMEOUT_SECONDS` default of `10` → `30`), and added a+ regression test (`services/agents/tests/test_attribution_service_url.py`).+## [7.4.0] — 2026-05-29Security-hardening and platform release. Folds in the May 27–29 hardening wave,diff --git a/docs/threat-actor-attribution.md b/docs/threat-actor-attribution.mdindex 8819eee7f..0c69bf764 100644--- a/docs/threat-actor-attribution.md+++ b/docs/threat-actor-attribution.md@@ -65,7 +65,7 @@ There is **no** synthetic "every observed IOC counts as a match" inflation.## APIMounted under `/api/v1/actors` on the `threatintel` service (default-port `8083`).+port `8005`).### `POST /api/v1/actors/attribute`@@ -154,8 +154,8 @@ and `severity`. To enrich attribution, populate those fields on theincoming alert.The HTTP timeout is tunable via `AISOC_ATTRIBUTION_TIMEOUT_SECONDS`-(default `10`), and the `threatintel` base URL via `AISOC_THREATINTEL_URL`-(default `http://threatintel:8083`).+(default `30`), and the `threatintel` base URL via `AISOC_THREATINTEL_URL`+(default `http://threatintel:8005`).## v0 limits and intended next stepsdiff --git a/services/agents/app/agents/investigation_agent.py b/services/agents/app/agents/investigation_agent.pyindex 262a31f07..841e13afb 100644--- a/services/agents/app/agents/investigation_agent.py+++ b/services/agents/app/agents/investigation_agent.py@@ -20,7 +20,7 @@ logger = structlog.get_logger()# Address of the threat intel service that hosts the attribution endpoint.# Defaults to the in-cluster Docker Compose hostname; override via env in# Kubernetes / Fly / standalone deployments.-THREAT_INTEL_SERVICE_URL = os.getenv("AISOC_THREATINTEL_URL", "http://threatintel:8083").rstrip("/")+THREAT_INTEL_SERVICE_URL = os.getenv("AISOC_THREATINTEL_URL", "http://threatintel:8005").rstrip("/")ATTRIBUTION_TIMEOUT_SECONDS = float(os.getenv("AISOC_ATTRIBUTION_TIMEOUT_SECONDS", "30.0"))diff --git a/services/agents/tests/test_attribution_service_url.py b/services/agents/tests/test_attribution_service_url.pynew file mode 100644index 000000000..350b1f755--- /dev/null+++ b/services/agents/tests/test_attribution_service_url.py@@ -0,0 +1,108 @@+"""Regression tests for the threat-actor attribution client URL.++The investigation agent calls the ``threatintel`` service's+``POST /api/v1/actors/attribute`` endpoint. The ``threatintel`` service+binds port **8005** everywhere it is actually run — ``Dockerfile``+(``EXPOSE 8005`` / ``--port 8005``), ``docker-compose.yml``+(``127.0.0.1:8005:8005``), and the canonical service/port table in+``README.md``. An earlier default of ``http://threatintel:8083`` meant the+agent → attribution call could never connect: it has been pointed at a+port nothing listens on since the feature shipped.++These tests lock the contract in two places:++* the default base URL resolves to ``:8005`` (the regression guard for the+ ``8083`` typo), and an operator override via ``AISOC_THREATINTEL_URL`` is+ honoured with the trailing slash stripped;+* ``_call_attribution_service`` posts to ``{base}/api/v1/actors/attribute``+ so neither the port nor the path can silently drift again.++We mock the network with :mod:`respx` so the suite stays offline-safe,+mirroring ``test_fusion_client.py`` which guards the analogous fusion+``/process`` path/port mismatch.+"""++from __future__ import annotations++import importlib+import sys+from pathlib import Path++import httpx+import pytest+import respx++# Match the path-mutation pattern used by the rest of services/agents/tests/.+_AGENTS_ROOT = Path(__file__).resolve().parents[1]+if str(_AGENTS_ROOT) not in sys.path:+ sys.path.insert(0, str(_AGENTS_ROOT))+++def _reload_agent(monkeypatch: pytest.MonkeyPatch, url: str | None):+ """Reload ``investigation_agent`` under a controlled env.++ The module reads ``AISOC_THREATINTEL_URL`` at import time, so the env+ must be set/cleared before the reload. We use ``importlib.import_module``+ (rather than ``import app.agents.investigation_agent as ...``) so+ CodeQL's py/import-and-import-from rule doesn't flag this file.+ """+ if url is None:+ monkeypatch.delenv("AISOC_THREATINTEL_URL", raising=False)+ else:+ monkeypatch.setenv("AISOC_THREATINTEL_URL", url)+ module = importlib.import_module("app.agents.investigation_agent")+ return importlib.reload(module)+++def test_default_base_url_uses_canonical_port(monkeypatch: pytest.MonkeyPatch) -> None:+ """With no override, the agent targets ``threatintel:8005`` — never ``:8083``.++ 8005 is where the threatintel service actually listens (Dockerfile,+ docker-compose, README service table). A regression to ``:8083`` would+ silently break threat-actor attribution end to end.+ """+ module = _reload_agent(monkeypatch, None)+ assert module.THREAT_INTEL_SERVICE_URL == "http://threatintel:8005"+ assert ":8083" not in module.THREAT_INTEL_SERVICE_URL+++def test_env_override_is_honoured_and_trailing_slash_stripped(+ monkeypatch: pytest.MonkeyPatch,+) -> None:+ """``AISOC_THREATINTEL_URL`` wins for non-Compose deployments."""+ module = _reload_agent(monkeypatch, "http://ti.internal:9000/")+ assert module.THREAT_INTEL_SERVICE_URL == "http://ti.internal:9000"+++async def test_call_attribution_service_posts_to_correct_url(+ monkeypatch: pytest.MonkeyPatch,+) -> None:+ """The client must POST to ``{base}/api/v1/actors/attribute``.++ Guards both the port and the path together, so neither can drift back+ out of sync with where the threatintel router is mounted.+ """+ base = "http://threatintel-test:8005"+ module = _reload_agent(monkeypatch, base)++ attribution = {+ "actor_id": "APT28",+ "actor_name": "APT28 (Fancy Bear)",+ "confidence_score": 0.5,+ "reasoning": ["Matched 2/3 TTPs"],+ }++ async with respx.mock(base_url=base, assert_all_called=True) as router:+ route = router.post("/api/v1/actors/attribute").mock(+ return_value=httpx.Response(200, json=attribution),+ )++ result = await module._call_attribution_service(+ iocs=[{"value": "evil.exe", "type": "filename"}],+ mitre_techniques=["T1566", "T1059"],+ case_metadata={"targets": ["government"]},+ )++ assert result == attribution+ assert route.called+ assert str(route.calls.last.request.url) == f"{base}/api/v1/actors/attribute"